Coding & Dev Tools

dikamilo Skill

Mine the project's real conventions from tooling, code, and docs and write them into the standards layer.

fw1201 Bundle

產出可本機開啟的互動教學網頁。適用於互動測驗、學習卡、抽籤、計時器。

fw1201 Bundle

設計驅動問題、探究歷程與真實成果。適用於PBL、專題式學習。

fw1201 Bundle

建立與更新可供技能讀取的教師偏好。適用於教師設定、設定教學偏好。

fw1201 Bundle

依學習證據調整任務與支持。適用於差異化、分層任務、UDL。

fw1201 Bundle

製作有答案、解析與配分的評量。適用於出題、試卷、素養題。

fw1201 Bundle

根據作品證據撰寫具體可行的回饋。適用於評語、作業回饋。

fw1201 Bundle

依課程目標安排活動、時間與評量。適用於教案、備課、108課綱。

fw1201 Bundle

建立任務專屬的表現描述與評分方式。適用於rubric、評量規準。

fw1201 Bundle

整理計畫、會議與行政文稿。適用於公文、行政文書、活動計畫。

fw1201 Bundle

設計共同規範、班級活動與支持策略。適用於班級經營、班規、班級文化。

fw1201 Bundle

跨單元安排學期目標、進度與課綱對應。適用於課程地圖、學期規劃。

fw1201 Bundle

整合不同學科的概念與探究任務。適用於跨領域、跨科。

fw1201 Bundle

編排學生可完成的練習與思考任務。適用於學習單、練習單。

fw1201 Bundle

檢視評量證據與改善任務設計。適用於抗AI評量、評量真實性。

fw1201 Bundle

協助整理學習證據、反思與成果。適用於學習歷程、成果反思。

fw1201 Bundle

收集課中證據並決定教學調整。適用於形成性評量、出口票、提問。

fw1201 Bundle

撰寫清楚、有同理心的親師草稿。適用於親師溝通、家長信、聯絡簿。

henry-md Bundle

Surgically stage only the current chat's relevant changes while leaving unrelated work untouched. Use when the user explicitly invokes `add` to rebuild the index for this conversation only, including earlier chat context. By default it stages the chat-scoped result and suggests a commit message. `/add push` additionally commits only that work plus the minimum already-modified diff required for a successful Railway build, pushes checked-out `main`, then watches the resulting Railway deployment. If Railway fails for a generated or build-support artifact reason that can be fixed without editing developer-authored source files, regenerate the needed artifacts, amend the just-pushed commit, and force-push with lease instead of creating a separate fix commit. This skill never creates branches or opens PRs.

henry-md Bundle

Show the raw contents of another skill's SKILL.md file when the user explicitly invokes the see skill, such as `/see test` or `$see test`. If the user invokes see without a second argument, treat it as a request to inspect the most recently created skill in the current conversation history. Do not use for general code or file browsing.

henry-md Bundle

Capture and inspect a page from a reusable signed-in browser session, then take a real screenshot. Default to an anonymous per-thread local account so `/check` runs do not interfere with each other. Only use Henry Deutsch's real account when the user explicitly asks for Henry-specific data or asks to inspect `henrymdeutsch@gmail.com`.

henry-md Bundle

Keep Claude and Codex skills in feature parity by comparing `~/.claude/skills` and `~/.codex/skills`, finding missing skills on either side, and reconciling functionality gaps inside matching `SKILL.md` files. Treat matching Codex built-ins such as `skill-creator` as satisfying the Codex side so duplicate custom Codex files are not created. Use when the user wants to sync custom commands across Claude and Codex, mirror one or more named skills between them, or close behavior gaps without copying the entire built-in Codex catalog into Claude.

henry-md Bundle

First reproduce and screenshot the reported UI or product issue, then iteratively verify, stress, and fix it by using screenshots and direct browser inspection to generate bug hypotheses, probe likely breaking routes or states, edit code for confirmed issues, and loop until the feature is visibly right and no obvious adjacent breakage remains. Use when the user invokes `/checks`, wants you to keep checking a page until it looks right, or expects an autonomous bug-hunting, screenshot, compare, edit, and fix loop rather than a single passive verification pass.

henry-md Bundle

Perform a thorough TypeScript cleanup pass when the user explicitly invokes the fix-ts skill to address compiler failures, high-value lint issues, or build-breaking type regressions. Use it for TypeScript-heavy cleanup work where errors should be grouped by root cause and fixed cleanly at the source rather than patched with broad casts or one-off assertions.

henry-md Skill

Investigate a failed Railway deployment when the user explicitly invokes the railway skill or asks to debug a Railway deployment failure. Anchor the work to a specific failed deployment ID, read the full deployment-specific logs, reproduce the failing Railway command locally, make the smallest relevant fix, verify locally, stage only the Railway-related changes without committing or pushing, and self-improve this skill when reusable learnings would save significant time in future Railway investigations.

henry-md Skill

Open a route in a real browser with Playwright, capture a screenshot, and visually inspect the result before claiming the page works. Use when the user asks you to see a page, verify that a route loads, or check UI changes directly.

henry-md Bundle

Capture and compare visual screenshots between a public reference URL and a local or deployed candidate URL. Use when cloning an existing site pixel-for-pixel, validating responsive similarity, generating screenshot heatmaps, comparing public-vs-local pages, auditing exactness across desktop/mobile/tablet screen sizes, or showing the user visual diff artifacts.

henry-md Bundle

Reviews all staged, unstaged, and untracked changes in the working tree and groups them into logical buckets that likely belong in separate commits or parallel tasks. Use when the user wants to understand a dirty tree before staging or committing, split local changes into coherent chunks, or explicitly asks for a `push` mode that stages and commits each bucket locally, verifies the stack, then pushes once.

henry-md Bundle

Review an existing Codex skill for structure, verbosity, triggering quality, and overall fit with Codex skill best practices. Use when the user explicitly invokes `/skill-doctor` with an existing skill name, or otherwise asks for a serious skill health check before rewriting a skill. This skill is review-only on the first pass and requires explicit follow-up consent before editing any files.

henry-md Bundle

Reference and diagnostic guidance for Cursor/VS Code configuration, workspace state, and extension issues. Use when the user asks why Cursor is slow, hung, or unable to load terminals, diffs, Git/SCM providers, GitLens views, or source control state; when Cursor says "No source control providers registered"; when raw Git may be fast but Cursor UI is stuck; or when preserving learnings about Cursor settings, logs, workspaceStorage, terminal persistence, and repo-specific cache resets.

henry-md Bundle

Query aggregate Firebase Auth, Firestore, sharing, billing, and tracked-usage stats for the Intention Setting app. Use when the user asks about Firebase access, user counts, active users, app usage stats, public share links, subscriptions, or other user metrics for /Users/Henry/Developer/intention-setting.

yansircc Bundle

Use when you need to run, inspect, or participate in roundtable-kernel sessions through the bundled `rtk` runtime. Covers autonomous runs, live chair/critic handoff, durable session truth, and common commands such as `help`, `show`, `wait`, `next`, and `apply`.

ysyecust Skill

生产就绪的 Dart 和 Flutter 模式，涵盖空安全、不可变状态、异步组合、Widget 架构、流行的状态管理框架（BLoC、Riverpod、Provider）、GoRouter 导航、Dio 网络请求、Freezed 代码生成和整洁架构。

ysyecust Skill

日本語翻訳：このファイルは quality-nonconformance 用の日本語翻訳が必要です

zhaji2333 Skill

当目标存在支付/下单/退款/提现/转账/优惠券/积分/红包/会员/订阅/审批/库存/抽奖等业务功能，或发现状态可跳变、金额参数可控、并发可重放时调用。负责业务状态机建模、金额篡改、订单状态跳变、竞态条件与重放攻击深度挖掘。

zhaji2333 Skill

当发现XML导入/SOAP/Office解析、序列化数据（Java/PHP/Python/.NET）、JSON深合并、yaml.load、不可信反序列化入口时调用。负责反序列化RCE链、XXE、原型污染、phpggc/ysoserial利用链构造与验证。